1 / 21

The anYnt Project Intelligence Test  one

The anYnt Project Intelligence Test  one. Javier Insa-Cabrera 1 , José Hernandez-Orallo 1 , David L. Dowe 2 , Sergio España 1 , M.Victoria Hernandez-Lloreda 3 ,. Departament de Sistemes Informàtics i Computació , Universitat Politècnica de València , Spain .

Download Presentation

The anYnt Project Intelligence Test  one

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The anYnt Project Intelligence Test one Javier Insa-Cabrera1, José Hernandez-Orallo1, David L. Dowe2, Sergio España1, M.Victoria Hernandez-Lloreda3, Departament de SistemesInformàtics i Computació, UniversitatPolitècnica de València, Spain. Computer Science & Software Engineering, Clayton School of I.T., Monash University, Clayton, Victoria, 3800, Australia. Departamento de Metodología de lasCiencias del Comportamiento, Universidad Complutense de Madrid, Spain CQRW2012 - AISB/IA-CAP 2012 World Congress, July 4-5, Birmingham, UK

  2. Measuring intelligence universally • Precedents • oneTest setting • Testing AI performance • Testing different systems • Discussion Outline

  3. Measuring intelligence universally • Can we construct a ‘universal’ intelligence test? • Project: anYnt (Anytime Universal Intelligence) • http://users.dsic.upv.es/proy/anynt/ • Any kind of system (biological, non-biological, human). • Any system now or in the future. • Any moment in its development (child, adult). • Any degree of intelligence. • Any speed. • Evaluation can be stopped at any time.

  4. Precedents • Imitation Game “Turing Test” (Turing 1950): • It is a test of humanity, and needs human intervention. • Not actually conceived to be a practical test for measuring intelligence up to and beyond human intelligence. • CAPTCHAs (von Ahn, Blum and Langford 2002): • Quick and practical, but strongly biased. • They evaluate specific tasks. • They are not conceived to evaluate intelligence, but to tell humans and machines apart at the current state of AI technology. • It is widely recognised that CAPTCHAs will not work in the future (they soon become obsolete).

  5. Precedents • Tests based on Kolmogorov Complexity (compression-extended Turing Tests, Dowe 1997a-b, 1998) (C-test, Hernandez-Orallo 1998). • Look like IQ tests, but formal and well-grounded. • Exercises (series) are not arbitrarily chosen. • They are drawn and constructed from a universal distribution, by setting several ‘levels’ for k: • However... • Some relatively simple algorithms perform well in IQ-like tests (Sanghi and Dowe 2003). • They are static (no planning abilities are required).

  6. Precedents • Universal Intelligence (Legg and Hutter 2007): an interactive extension to C-tests from sequences to environments. = performance over a universal distribution of environments. • Universal intelligence provides a definition which adds interaction and the notion of “planning” to the formula (so intelligence = learning + planning). • This makes this apparently different from an IQ (static) test. μ π oi ri ai

  7. Precedents • KolmogorovComplexity where l(p) denotes thelength in bits of p and U(p) denotes theresult of executing p on U. • Universal Distribution Given a prefixed-free machine U, the universal probability of string x isdefined as:

  8. Precedents • Levin’sKtComplexity where l(p) denotes thelength in bits of p and U(p) denotes theresult of executing p on U, and time(U,p,x) denotes the time that U takesexecuting p to produce x. • Time-weighted Universal Distribution Given a prefix-free machine U, the universal probability of string x isdefined as:

  9. Precedents • A definition of intelligence does not ensure an intelligence test. • Anytime Intelligence Test (Hernandez-Orallo and Dowe 2010): • An interactive setting following (Legg and Hutter 2007) which addresses: • Issues about the difficulty of environments. • The definition of discriminative environments. • Finite samples and (practical) finite interactions. • Time (speed) of agents and environments. • Reward aggregation, convergence issues. • Anytime and adaptive application. • An environment class  (Hernandez-Orallo 2010).

  10. oneTest setting • Discriminativeenvironments. • Interactinfinitely: Must be a pattern (Good and Evil). • Balancedenvironments. • Symmetricrewards. • SymmetricbehaviourforGood and Evil. • Agentshaveinfluenceonrewards: Sensitivetoagents’ actions.

  11. oneTest setting • Implementation of the environment class: • Spaces are defined as fully connected graphs. • Actions are the arrows in the graphs. • Observations are the ‘contents’ of each edge/cell in the graph. • Agents can perform actions inside the space. • Rewards: Two special agents Good (⊕) and Evil (⊖), which are responsible for the rewards.

  12. Testing AI performance • Test with 3 different complexity levels (3,6,9 cells). • We randomly generated 100 environments for each complexity level with 10,000 interactions. • Size for the patterns of the agents Good and Evil (which provide rewards) set to 100 actions (on average). • Evaluated Agents: • Q-learning • Random • Trivial Follower • Oracle

  13. Testing AI performance • Experiments with increasing complexity. • Results show that Q-learning learns slowly with increasing complexity. 3 Cells 6 Cells 9 Cells

  14. Testing AI performance • Analysis of the effect of complexity: • Complexity of environments is approximated by using (Lempel-Ziv) LZ(concat(S,P)) x |P|. 9 Cells Allenvironments • Inverse correlation with complexity (difficulty , reward ).

  15. Testing different systems • Each agent must have an appropriate interface that fits its needs (Observations, actions and rewards): • AI agent • Biological agent: 20 humans b:E:πGa:: +1.0

  16. Testing different systems • We randomly generated only 7 environments for the test: • Different topologies and sizes for the patterns of the agents Good and Evil (which provide rewards). • Different lengths for each session (exercise) accordingly to the number of cells and the size of the patterns. • The goal was to allow for a feasible administration for humans in about 20-30 minutes.

  17. Testing different systems • Experiments were paired. • Results show that performance is fairly similar.

  18. Testing different systems • Analysis of the effect of complexity : • Complexity is approximated by using LZ (Lempel-Ziv) coding to the string which defines the environment. • Lower variance for exercises with higher complexity. • Slight inverse correlation with complexity (difficulty , reward ).

  19. Discussion • Environment complexity is based on an approximation of Kolmogorov complexity and not on an arbitrary set of tasks or problems. • So it’s not based on: • Aliasing • Markov property • Number of states • Dimension • … • The test aims at using a Turing-complete environment generator but it could be restricted to specific problems by using proper environment classes. • An implementation of the Anytime Intelligence Test using the environment class  can be used to evaluate AI systems.

  20. Discussion • The test is not able to evaluate different systems and put in the same scale. The results show this is not a universal intelligence test. • What may be wrong? • A problem of the current implementation. Many simplifications made. • A problem of the environment class. • A problem of the environment distribution. • A problem with the interfaces, making the problem very difficult for humans. • A problem of the theory. • Intelligence cannot be measured universally. • Intelligence is factorial. Test must account for more factors. • Using algorithmic information theory to precisely define and evaluate intelligence may be insufficient.

  21. Thank you! • Some pointers: • Project: anYnt (Anytime Universal Intelligence) • http://users.dsic.upv.es/proy/anynt/ • Havefunwiththe test. • http://users.dsic.upv.es/proy/anynt/human1/test.html

More Related